2:5 Design for Cyclic Loading

1. Completely reversing cyclic stress and endurance strength Cyclic loading has already been introduced in the context of stress concentration. A purely reversing or cyclic stress means when the stress alternates between equal positive and negative peak stresses during each cycle of operation. Cyclic stress over time can be represented by a sinusoidal curve as shown. In this diagram the stress varies with time Pure cyclic stress between +250 MPa to -250MPa. I n pure cyclic stress the average stress is zero. As 300 has been stated earlier, this kind of stress can 200 arise in rotating members under bending.
Stress (MPa) 100 0 -100 -200 -300 time

When a part is subjected to cyclic stress, also known as range stress (Sr), it has been observed that the failure of the part occurs after a number of stress reversals (N) even it the magnitude of the range stress is below the materials yield strength. Generally, higher the range stress, lesser number of reversals is needed for failure.

No. of stress reversals (N) 1000 2000 4000 8000 16000 32000 64000 96000 144000 216000 324000 486000 729000 1000000

Max cyclic stress (Sr) psi 81000 75465 70307 65501 61024 56853 52967 50818 48757 46779 44881 43060 41313 40000

For a typical material, the table and the graph above show the relationship between the magnitudes of range stress (Sr in psi) and the number of stress reversals (N) before failure of the part. For example, if the part were subjected to Sr= 81,000 psi, then it would fail after N=1000 stress reversals. If the same part is subjected to Sr = 61,024 psi, then it can survive up to

Sengupta

MET 301: Design for Cyclic Loading

1 of 8

N=16,000 reversals. Thus, as we increase the number of stress reversals, the maximum cyclic stress a part can take gradually decreases. It has been observed that for most of engineering materials, the rate of reduction of Sr becomes negligible near the vicinity of N = 106 and the slope of the S-N curve becomes more or less horizontal. For the curve shown above, at N= 106 , the slope of the curve has become horizontal at Sr=40,000 psi. Because the slope of the above S-N curve is horizontal at N=106 reversals, that means if we keep the cyclic stress value less than Sr = 40,000 psi, then the part will survive indefinitely (infinite number of stress reversals). The maximum completely reversing cyclic stress that a material can withstand for indefinite (or infinite) number of stress reversals is known as the fatigue strength or endurance strength (Se) of the part material. This is essentially the max completely reversing cyclic stress that a material can withstand for N=106 , as the curve is horizontal after this point. For the example S-N curve shown above, Se = 40,000 psi. 2. Factors affecting fatigue strength The failure caused by cyclic stress is called fatigue failure . The fatigue failure originates from a micro-crack (brittle failure) due to stress concentration effect at surface irregularities. Minute irregularities such as grinding scratches, tool marks, inspection stamping, stamped part numbers or surface scales may produce a high value for the stress due to stress concentration and serve as the starting point for the micro crack, which progressively increases until the part breaks into two pieces similar to a brittle fracture. No plastic deformation is observed in the failure surface. Most commonly the fatigue failure occurs due to reversing bending stress in rotating machineries, less commonly due to reversing torsional shear stress and rarely from reversing axial stress. As the most highly stressed points are on the outer surface for both bending and torsion, crack originates on the outer surface and progresses inwa rdly. As a consequence of this, a weak decarburized layer on the outer surface of heat-treated parts often produce low endurance limit. Since fatigue cracks are due to tensile stress, a residual tensile stress on the surface of the part constitutes a fatigue hazard. Suc h a residual tensile stress, for example may arise from a cold working operation on the part without stress relieving . Parts that are finished by grinding frequently have a thin surface layer with high residual stress in tension. Such residual stress comb ined with the tensile stress from loading, may give a resultant stress sufficiently great to cause a fatigue crack to start. Any residual tensile stress on the surface should be removed, or better still, converted into a layer of compression. Shot blasting, or shot peening operations can induce residual compressive residual stress on surface. Also nitrided and carburised parts have residual compressive stress on the surface. Residual compressive stress on the surface will reduce the tensile stress from the load, and thus helps to improve fatigue characteristics of the part. Sand blasting of the part should be avoided since the scratches serve as stress risers. The fundamental measure of fatigue performance for a material is the endurance limit from a plain polished specimen. Because the brittle nature of the failure, the endurance limit of a

Sengupta

MET 301: Design for Cyclic Loading

2 of 8

material is closely related to the ultimate tensile strength (Su ) rather than yield strength (Syp ). See Figure 2-26 in textbook, which provides the relationships between the endurance limits and the ultimate tensile strengths of steel specimens with different surface finishes. Typically for wrought steels when the surface is ground and polished Se = 0.5 Su, for machined surface, Se = 0.35 to 0.4 Su and for as-forged or as-rolled surface Se = 0.2 Su. Corrosion of surface from water or acid may reduce the endurance strength to further lower value. 3. Design for fatigue stress Unlike a pure range stress (Sr) discussed above, a machine part may be subjected to a combined steady and range stress. Following design procedure handles such combined stress situation. A generalized one dimensional stress condition, can be defined as combine purely reversing stress or range stress (Sr) superimposed on a steady stress (Savg). The following stresstime graph shows this combined range and steady stress condition. If the stress is varying between Smax & Smin, then the

S max + S min 2 S max S min Range stress = S r = 2 If a part only contains the steady part of the stress S avg , then to prevent failure:
Steady stress = S avg = K*Savg < Syp/Nfs , when the range part of stress Sr = 0 Here, K= geometric stress concentration factor Syp = yield strength of the material Nfs= factor of safety

Usually parts subjected to fatigue loading are made of ductile material, and in this case the geometric stress concentration factor K=1. Thus for one extreme combination of stress, when there is only the steady part of the load, for safe design: (i) Savg < Syp/N fs & Sr = 0 Which means that Savg can go up to Syp/N fs when Sr = 0 Similarly, for the other extreme combination of stresses, when the stress is purely cyclic, then for safe design: (ii) Sr < Se /(Nfs *Kf) & Savg =0 Which means Sr can go up to Se /(Nfs *Kf), when Savg =0 (Kf = fatigue stress concentration factor, Se= endurance strength of the material, Nfs= factor of safety) Sengupta MET 301: Design for Cyclic Loading 3 of 8

If we plot steady stress (Savg) along x axis and the range stress (Sr) along y axis, then the two extreme stress conditions (i) & (ii) described above, constitute two point on x and y axis. Soderberg Line is obtained by joining these two points . When in a machine part, both types of stress are present simultaneously, if the stress combination (Savg & Sr) is contained in the blue area defined by the Soderberg line, then the part should be safe. Any stress combination falling above the Soderberg line can be considered unsafe. Using intercept form of the equation of straight line, ie., x/a+y/b=1, the safe design area (blue area) can be defined by: S avg Sr + 1 S yp Se N fs K f N fs
Multiplyin g both side by S yp N fs

S yp S yp S avg + S r K f S N .....................(1) e fs Equation (1) is called Soderberg Equation for design of a part with combined steady and range stress. S Note that, the right hand side of the equation yp , which is the design limit for normal N fs

S yp steady stress Savg . Because of the presence of the range stress Sr, the factor Sr K f S is e added, which is the static equivalent of the range stress Sr. This information will be used later in the design of shaft in Chapter 3 in the textbook.
Goodman proposed that the Soderberg design limits are conservative, and he extended the safe design stress for steady stress from Syp /N fs to Su/N fs. This is also due to the assumption that fatigue failure occurs in brittle failure mode. This resulted in the safe design space as shown and the resulted in Goodman Design equation:

Su S avg + S r K f S e

Su N .........( 2) fs

Goodman Equation can be obtained from Soderberg equation by replacing Syp by Su .

Sengupta

MET 301: Design for Cyclic Loading

4 of 8

However, in the safe area defined by Goodman line, when the magnitude of steady stress Savg becomes more than Syp /Nfs, the part may fail from yielding from plastic deformation. The area is shown as unsafe region. To eliminate this shortcoming, modified Goodman equation proposed the shaded area as the safe design space. This space is obtained by drawing a line with 45o angle from the Syp /N fs point on the x axis. Mathematically this Modified Goodman space is equivalent to satisfying the following two equations (3) & (4), simultaneously.
Su Su S avg + S r K f .........(3) S e N fs S yp S avg + S r K f .................( 4) N fs

4. Design for finite life All of the above design equations, the upper limit of the cyclic stress is kept below the endurance strength Se, of the part material. Theoretically Se is the max range stress Sr at N = 106 . We have seen that the S-N curve becomes horizontal after N = 106 , which means that if a part survives 106 stress alterations, it will survive for ever (indefinitely). In some design situations however, the number of stress reversals a part is designed undergo is limited by the life of the part. For example, the rotating parts in a booster rocket may be designed to perform for merely several thousands of rotations, because the booster will be discarded after few hours of the flight. Obviously, in such a situation, we can use a higher Sr value to support the desired number of rotations, and dont have to design for indefinite life. Thus, when number of stress reversals is less than 106 , we can work with a higher Sr value and save some material. For example, from the above S-N curve, if the design has to survive up to 100,000 stress reversals, then, the part can be loaded up to a range stress level of Sr 50,000 psi. For indefinite life, Sr is limited to Se= 40,00 psi.

Sengupta

MET 301: Design for Cyclic Loading

5 of 8

Basquin has proposed that the equation for S-N curve which is true for many materials. The Basquin equation can be given by: S r = AN B , where Sr is a reversing stress, which is more than Se and which will give a fatigue life of N cycles. A & B are two material constants obtainable by the method described below. Taking log : log( S r ) = log A + B log( N )........(5) For a small number of stress reversals, N = 1000, the range stress, Sr can be increased to 90% of the ultimate tensile strength of the material, or, Sr = 0.9Su for N = 1000. Putting this pair of values in (5), we get log( 0.9S u ) = log A + B log( 1000) or , log( 0.9S u ) = log A + 3B...........................( 6) We also know that for N= 106 , Sr can go up to Se, that is, Sr = Se, for N = 106 . Putting values in (5) we get log( S e ) = log A + B log( 1000,000) or , log( S e ) = log A + 6 B...........................( 7) Subtracting (7)-(6) and simplifying: B = From (7): A =

log( S e ) log( 0.9S u ) 3

Se 10 6B Thus from the known values of ultimate tensile strength (Su) and the endurance strength (Se) of the part material, we can determine the material constants A and B of Basquins equation. Once the constants A & B are known, then we can determine the maximum completely reversing range stress Sr for any value of N less than 106 using Basquins equation Sr = AN B
Also, when the constants A & B are known, then for a given completely reversing stress Sr greater than Se, we can determine the number of stress reversals (N) after which fatigue failure
S B will occur : N = r . A
1

5. Design for finite life with combined steady (Savg ) and range stress (Sr) Basquins equation is applicable for a completely reversing range stress. For a combined steady (Savg) and range (Sr) stress, if we want to apply Basquins equation, we first need to determine the equivalent range stress (SR) from the combined stress situation. Once the equivalent range stress (SR) is determined, then Basquins equation can be used to obtain the number of reversals N for this equivalent SR.

Sengupta

MET 301: Design for Cyclic Loading

6 of 8

Let us say, calculated steady and range stresses for a part are Savg & Sr, respectively. The actual steady and range stress becomes Savg & KfSr, respectively after considering stress concentration factor. Kf is the fatigue stress concentration factor applicable to the range stress. We know that for ductile material, geometric stress concentration factor K = 1, hence the actual steady stress remains unchanged from the calculated stress. The equivalent range stress SR from the actual combined stress (Savg & Kf Sr) is determined from Goodman line shown (without Nfs). We know that a combined stress denoted by a point below the Goodman line (in blue shaded area), will provide indefinite life. For finite life situation, the applied stress point M (Savg, KfSr) must be above the Goodman line. To obtain the equivalent range stress, we draw a line from Su on x axis to the point M and extend the line to y axis. The point SR is the intersection of this line to y axis, which denotes the equivalent range stress. The magnitude of SR can be easily obtained K f Sr S from the two similar triangles OSuSR and SavgMSu : R = S u S u S avg K f S r Su or , S R = S u S avg Thus, for designing with finite life of a part (N<106 ), the applied Savg & Sr can be converted to an equivalent SR and then Basquins equation can be applied. Also note that, the ultimate tensile strength (Su ) for the material is needed to find this equivalent range stress. 6. Miners Equation for Finite Life Design A machine part may have one stress for a portion of its life, another stress for another portion, and so on. Each combination of Savg & Sr, can be reduced to equivalent completely reversed stresses SR1 , SR2 , SR3 and so on. Let fatigue life for stress SR1 , if applied alone is N1 reversals, and let n1 be the actual number of reversals at stress level SR1 . Then stress SR1 has consumed n1 /N 1 portion of the life of the part. Similarly, let N2 be the fatigue life at stress SR2 , and let n2 be the actual reversals occurred at this stress and thus consumed n2 /N 2 portion of life. Sum of these proportions constitutes the entire life, or unity:

n1 n n + 2 + 3 + ..... = 1 N1 N 2 N3
The above equation is called Miners Equation. Using this equation we can determine the amount of life remaining in a part at any stress level, if we know the past history of its service.

Sengupta

MET 301: Design for Cyclic Loading

7 of 8

7. Concluding Remark 1. As you have notices that all the theories about cyclic stress developed here pertain to uniaxial stress. For biaxial or triaxial stresses , we know the stresses may interact differently, and prediction of failure is not straight forward. In case of cyclic stress one more problem arises, that is the cyclic stresses in different axes may have different frequenc ies. Design procedure for such stresses is beyond the scope of this course. 2. Failure with fluctuating stresses that are high enough to cause failure in a few thousand cycles or less is called low-cycle fatigue . Plastic yielding at localized areas may be involved. The theory is beyond the scope of this course.

Sengupta

MET 301: Design for Cyclic Loading

8 of 8